Flaxseed, grape seed and black cumin seed CPO were obtained from a local cold-pressing company (Oneva Cold Pressing, Istanbul, Turkey). Chemicals and solvents with the highest purity available were used without further purification.

CD and PV values of CPO during thermal oxidation at 60 °C were measured according to AOCS methods Cd 8-53 and Cd 18-90 (AOCS, 1997).

Fatty acid methyl esters (FAME) were prepared according to IUPAC (1987). GC analysis was carried out with a Shimadzu GC-2010 chromatograph equipped with a fused capillary DB-23 fused-silica column (0.25 mm i.d., 60 m, 0.25 μm film thickness, Agilent J&W, USA). The carrier gas was helium at a flow rate of 0.70 mL/min. The column temperature was set to isothermal at 190 °C for 95 min, wherein the injector and detector temperatures were 230 °C and 240 °C, respectively. FAME peak areas were identified by the comparison of retention times with reference standards.

Aliquots of CPO were dissolved in n-hexane (5 mL) and mixed with 10 mL methanol: water (80:20, v/v) in a glass tube for two min in a vortex (Ramadan et al., 2012). After centrifugation for 10 min at 3000 rpm, the hydroalcoholic extract was separated from the lipid phase using a Pasteur pipette and then combined and concentrated in vacuo at 30 °C. The oily residue was re-dissolved in 10 mL methanol: water (80:20, v/v) and the extraction process was repeated three times. Hydroalcoholic extracts were re-dissolved in acetonitrile (15 mL) and the mixture was washed three times with n-hexane (15 mL each). Purified phenolics were concentrated in vacuo at 30 ˚C and then dissolved in methanol. The phenolic extracts (40 μL) and 30 mL of water were poured into a tube. 200 μL of Folin-Ciocalteu reagent were added to the mixture. After 7.5 min, 600 μL of sodium carbonate (38%) were mixed into the solution and the mixture was left for 180 min at room temperature. The absorbance was measured at 765 nm using a UV-VIS spectrophotometer (PG Instruments, England). Gallic acid was used as a standard and TPC was calculated and expressed as mg GAE/kg oil.

Phenolic and bioactive compounds were measured by reversed-phase high-performance liquid chromatography (RP-HPLC, Shimadzu Scientific Instruments, Tokyo, Japan). Identification and quantification were performed with a ShimadzuLC-10ADvp pump, a CTO-10Avp column heater, a Diode Array Detector, SCL-10Avp system controller, DGU-14A degasser and SIL-10ADvp auto-sampler (Shimadzu, Columbia, MD). Separations were performed at 30 °C on Agilent-Eclipse XDB C-18 reversed-phase column (4.6 mm length, 250 mm, 5 μm particle size). The mobile phases were A: 2.0% acetic acid in distilled water and B: methanol. The flow rate was 0.8 mL/min. For analysis, 25 mg of the extract was dissolved in 1 mL methanol wherein the injection volume was 10 μL. A modified gradient elution of two solvents was used: solvent A consisted of acetic acid: water (3:97, v/v), solvent B: methanol. The gradient program used is given in Table 1 according to Caponio et al., (2001). Identification and quantitative analyses were performed by comparison with standards. The limit of detection (LOD) and the limit of quantification (LOQ) were 0.01 and 0.02 μg/mL for gallic acid, 0.05 and 0.16 μg/mL for protocatechuic acid, 0.02 and 0.05 μg/mL for caffeic acid, 0.005 and 0.01 μg/mL for p-hydroxy benzoic acid, 0.02 and 0.07 μg/mL for chlorogenic acid, 0.01 and 0.02 μg/mL for syringic acid, 0.01 and 0.03 μg/mL for p-coumaric acid, 0.01 and 0.02 μg/mL for ferulic acid, 0.06 and 0.19 μg/mL for benzoic acid, 0.005 and 0.01 μg/mL for o-coumaric acid, 0.10 and 0,31 μg/mL for catechin, 0.05 and 0.15 μg/mL for (^-)-epicatechin, 0.03 and 0,10 μg/mL for vanilic acid, 0.01 and 0.04 μg/mL for vanilin, 0.06 and 0.18 μg/mL for rutin, 0.03 and 0.08 μg/mL for apigenin, 0.11 and 0.32 μg/mL for quercetin, 0.03 and 0.10 μg/mL for luteolin and 0.02 and 0.06 μg/mL for campherol, 0.73 and 2.21 μg/mL for thymoquinone, 2.49 and 7.56 μg/mL for thymol, 0.12 and 0.35 μg/mL for acacetin, respectively. The levels of phenolics in the extracts were calculated as mg/ kg oil. All extractions and chromatographic analyses were performed in triplicate and the results were averaged.

Tocopherols were analyzed according to the American Oil Chemists’ Society method Ce 8-89 (AOCS, 2003). Tocopherols (α, β, γ, and δ) were analyzed using normal-phase HPLC by the direct injection of CPO samples dissolved in a mixture (95:5, v/v) of heptane:tetrahydrofuran (THF). HPLC analysis was carried out with an SCL-10Avp System controller, LC-10ADvp pump, SIL-10ADvp Autosampler, CTO-10 Avp column heater and fluorescence detector set at 295 nm for excitation and 330 nm for emission. The 150 cm x 4.6 mm i.d. column filled with Supelcosil Luna, 5m (Supelco, Bellefonte, PA) was used. The mobile phase was a mixture of heptane:THF (95:5, v/v) at a flow rate of 1.2 mL/min wherein the injection volume was 10 mL. The standard samples used for identification and quantification were α, β, γ and δ-tocopherols (Sigma, St. Louis, Mo., USA). The data were analyzed and integrated using the Shimadzu Class-VP Chromatography Lab Automated Software. The levels of tocopherols in CPO were calculated as mg/kg oil using the external calibration curves obtained for each tocopherol standard. The LOD and the LOQ were 0.03 and 0.08 μg/mL for α-tocopherol, 0.02 and 0.06 μg/mL for β-tocopherol, 0.04 and 0.13 μg/mL for γ-tocopherol, and 0.03 and 0.08 μg/mL for δ-tocopherol. The chromatographic analysis was performed three times and the results were averaged.

CPO samples stored under accelerated oxidation conditions at 60 ºC were used to determine volatile compounds and VOC. Equilibration took place for 15 min at 35 °C followed by extraction (45 min) from the headspace at 35 °C then a 10 min of desorption in the GC injection port with the aid of a CTC Combi PAL (CTC Analytics AG, Zwingen, Switzerland) auto-sampler with 75 μm carboxen/polydimethylsiloxane solid-phase micro-extraction (SPME) fiber. GC-MS analyses were carried out with an Agilent system (GC 7890, MS 5975 N MSD, Santa Clara, CA, USA) using an analytical column of medium polarity (DB-624, 30 m length × 0.25 mm ID × 1.4 μm film thickness, Agilent Technologies, CA, USA). The following temperature program was applied: hold for 5 min at 40 °C; 3 °C/min up to 110 °C; 4 °C/min up to 150 °C; 10 °C/min up to 210 °C and hold for 12 min. The temperatures for the injection port, ion source, quadrupole, and interface were 250 °C, 230 °C, 150 °C, and 240 °C, respectively. Mass spectra were recorded in full scan mode at the electron impact of 70 eV with a scan range from m/z 41 to 400. The identification of volatile compounds was made by comparing mass spectra, Kovats index (KI) with the authentic standards and published data, as well as by comparing their mass spectra with Wiley7.0 (Wiley, NY, USA) and the mass spectrometry library of Nist 05 (National Institute of Standards and Technology, Gaithersburg, MD, USA). The KI parameters were calculated using the n-hydrocarbons (C4 to C20) series. One vial for daily assessment from sets of replicate vials for every CPO was used throughout the study.

Results are shown and expressed in terms of the mean and standard deviations. Significant differences were calculated using ANOVA in combination with Duncan’s test with a significance level of α=0.05.

The fatty acid profiles of the CPO samples are shown in Table 2. The main fatty acids in FSO were linolenic acid (56.5%), followed by oleic acid (18.0%) and linoleic acid (15.9%). The fatty acid compositions of GSO and BSO included linoleic acid (66.8% and 56.8%, respectively) as the major fatty acid, followed by oleic acid (19.5% and 23.8%) and palmitic acid (8.25% and 12.4%), respectively. The fatty acid composition of FSO was in agreement with those reported by others (Brühl et al., 2008; Ivanova-Petropulos et al., 2015; Raczyk et al., 2016; Teh and Birch, 2013). The results of the FAME profile of GSO fall within similar ranges reported (Tuberoso et al., 2007; Lutterodt et al., 2011; Wen et al., 2016). In addition, the fatty acid composition of BSO was in accordance with those previously reported in the literature (Cheikh-Rouhou et al., 2007; Lutterodt et al., 2010; Ramadan et al., 2012).

The TPC of CPO is given in Table 3. The highest TPC was found in BSO (1140 mg/kg oil), while FSO contained the lowest TPC (554.4 mg/kg). The TPC in BSO was similar to the values (1.02-1.40 mg GAE/g) reported by Lutterodt et al., (2010). The TPC of FSO was lower than the values reported by Kasote et al., (2013) who determined that the TPC of cold-pressed linseed oils ranged from 10 to 26 mg GAE/100 g oil. In addition, the TPC of cold-pressed GSO was similar to the values (0.16-0.80 mg GAE/g) reported by Lutterodt et al., (2011).

The phenolic and bioactive compound profiles of CPO are also given in Table 3. In the FSO, luteolin and apigenin were detected and their amounts were 0.76 and 0.02 mg/kg oil, respectively. In GSO, 8 compounds including vanillin, vanillic acid, p-coumaric acid, dihydroquercetin, luteolin, campherol, apigenin and acacetin were identified. GSO contained the highest dihydroquercetin content (2.49 mg/kg oil). Luteolin and apigenin were not detected in FSO or GSO in the research of Tuberos et al., (2007). The differences in bioactive and phenolic compound profiles could be due to differences in cultivar, variety, location and extraction conditions. The most abundant compound in black cumin oil was benzoic acid (228.5 mg/kg oil), followed by thymoquinone (97.58 mg/kg oil). The amounts of thymoquinone and benzoic acid in our study were higher than those reported by Kiralan et al., (2014) who determined that the levels of benzoic acid and thymoquinone were 4.15 and 14.4 μg/g, respectively.

Table 3 shows the tocopherol contents of CPO. γ-Tocopherol was found as the major tocopherol isomer in the FSO (137.9 mg/kg oil) and BSO (128.9 mg/kg oil). The main tocopherol in the GSO was the α form (123.0 mg/kg oil). The γ- tocopherol content in the FSO was in accordance with the values (10.5-15.0 mg/100 g seed) reported by Choo et al., (2007). The α-tocopherol value in the GSO in our work showed similarity to that determined by Fernandes et al., (2013) for oils of Portuguese GSO (85.5-244 mg/kg). The content of γ-tocopherol in the BSO was lower (0.225 g/kg) than that reported by Ramadan and Moersel (2004). The differences in the contents and composition of tocopherols in the CPO under study may be due to differences in cultivar, variety and origin of the investigated seeds.

The OSI values of CPO according to the Rancimat test revealed that BSO had the highest OSI (6.14 h), followed by GSO (4.06 h), while FSO had the lowest (2.63 h) OSI value. The results can be explained by the fatty acid profile of the CPO under study. The oxidative stability of FSO was very low in comparison to the other oils under Rancimat oxidation conditions. This might be explained by the fact that FSO is rich in linolenic acid, which has a high susceptibility to oxidation. Likewise, BSO had a better OSI value than GSO. GSO contained a higher level (66.8%) of linoleic acid than BSO (56.8%).

The OSI of FSO was higher than the OSI values (1.43-1.52 h) reported by Bozan and Temelli (2008). The results obtained for GSO heated at 110 °C were lower than those reported by Hassanien et al., (2014) for hexane-extracted GSO (8 h). However, the OSI of GSO was similar to those reported for refined GSO (3.3-5.3 h) under similar conditions (Hidalgo et al., 2002). The OSI value for BSO was lower than that (16.9 h) reported by Hassanien et al., (2014) and (ca. 22 h) Rudzińska et al., (2016), but higher than that reported by Kiralan et al., (2014) for cold-pressed BSO (3.48 h). These differences in OSI values might be due to the composition and amounts of natural antioxidants in the CPO samples.

The changes in PV in CPO during storage for 6 days at 60 °C are shown in Figure 1. The PV values for fresh CPO were 1.82, 12.2 and 26.5 for FSO, GSO and BSO, respectively. After 48 h of storage under oxidation conditions, the PV of FSO increased sharply and reached up to 29.97 meq O₂/kg. After that, PV values were recorded to show stabile behavior till the end of storage period. In addition, the PV values of GSO increased sharply after 48 h of storage and the highest values were recorded (81.6 meq O₂/kg). After 48 h of storage, the PV values for GSO slightly decreased to ca. 87-81 meq O₂/kg until the 5^th day of storage, when the PV of GSO decreased to 64 meq O₂/kg. In the BSO, the PV increased gradually with increasing storage period. The highest PV value (53.5 meq O₂/kg) for BSO was measured after 5 days of storage. Based on PV values, it could be concluded that BSO had the greatest oxidative stability among the CPO under study.

The changes in CD (K₂₃₂ values) of the CPO stored for 6 days under thermal oxidation are exhibited in Figure 2. The rate of increase in K₂₃₂ values showed similar behavior to that of the PV values. The K₂₃₂ values for FSO increased gradually with the increase in storage time, reaching the maximum value (6.41) at the end of the storage period. There was a significant increase in the K₂₃₂ values for the GSO with increasing storage time. In addition, a stable increase in the K₂₃₂ values of BSO was observed until the 6^th day of storage.

Our CD results for FSO are in agreement with Hamed and Abo-Elwafa (2012), who reported that the PV and CD values of FSO increased sharply when stored at 60 °C. Similar observations on PV and K₂₃₂ values of GSO stored under thermal conditions have been reported (Jang et al., 2015). In addition, similar increases in PV and CD values have been observed in BSO as affected by thermal oxidation conditions (Kiralan, 2014; Kiralan et al., 2017). However, our results might somehow differ from the results previously reported in the literature. In our experiment, small oil samples were used in headspace vials, as opposed to the use of large amount of oil samples in the above-mentioned literature. In addition, the differences in the results might be due to the distinctive experiment oxidation conditions, genotype and growing condition of plants.

An in-depth search in the scientific literature indicated that studying VOC in FSO, BSO and GSO during storage at thermal oxidation conditions had not yet been performed. The changes in the VOC and volatile compounds in CPO during storage under thermal oxidation conditions are given in Table 4. Six volatile compounds, including aldehydes and alkadienes, are detected in the headspace of FSO. During storage, hexanal was the major compound among VOC. 2,4-heptadienal, and (E,E)-2,4-heptadienal were formed upon oxidation and at the end of the storage period, both compounds increased up to the maximum values of 16.84 and 16.09x10⁶ AU, respectively. 2-hexenal, 2,4-hexadienal and E-2-heptenal were VOC detected after 3 days of FSO storage and their levels increased with the increase in storage time.

2,4-heptadienal was formed in canola and soybean oil with high linolenate (Snyder et al., 1985). (E,E)-2,4-heptadienal was also identified in soybean oil stored under thermal oxidation conditions (Snyder et al., 1988). Hexanal, 2-hexenal, (E)-2-heptenal, (E,E)-2,4-heptadienal and (E,E)-2,4-hexadienal were detected in fresh FSO without oxidation treatments (Krist et al., 2006; Wei et al., 2015). The results of our study on the VOC in FSO are in agreement with Abuzaytoun and Shahidi (2006) who reported an increase in hexanal levels during the storage of FSO at 60 °C.

Table 4 listed the VOC identified in the GSO during storage. Thirteen compounds were detected, including alkane (2-octene), aldehydes ((E)-2-heptenal, hexanal, (E)-2-octenal), ketones (3-octanone, 6-methyl-5-hepten-2-one, 2-heptanone, 2-octanone), and alkadienals (2,4-hexadienal, 2,4-heptadienal, 2,4-nonadienal, 2,4-decadienal, (E,E)-2,4-decadienal). (E)-2-heptenal was detected as the main VOC during storage, followed by hexanal. At the end of storage, the other most abundant aldehyde found in GSO was (E)-2-octenal. In addition, at the end of the storage experiment, diverse alkadienals were formed. However, the levels of these compounds remained limited among the VOC.

Hexanal, (E)-2-heptenal and (E)-2octenal were determined as the decomposition products of linoleate hydroperoxides (Frankel et al., 1981). The increase rate of forming hexanal and (E)-2-heptenal during storage at 60 °C was observed in the study by Jelen et at., (2000), who reported that the increase in 2-heptenal was more than that in hexanal in the headspace of cold-pressed rapeseed oil at the end of storage (60 °C). No literature report could be found about the volatile compounds of GSO during thermal oxidation but only one published work (Bail et al., 2008) reported on the volatile compounds of nine GSO. In this work (Bail et al., 2008), hexanal, 2-heptanone, (E)-2-heptenal and (E)-2-octenal were observed in the volatiles of oil samples. These compounds were also identified in our analyzed cold-pressed grape seed oil.

Five VOC were identified in the headspace of BSO during storage as given in Table 4. Most of the identified volatiles were aldehydes and only one ketone (2-heptanone) was detected in the headspace of the BSO sample. Among the aldehydes, hexanal and (E)-2-heptenal were formed as the main VOC during storage. Lower contents of 2-hexenal and (E)-2-octenal in the headspace of BSO were identified. The study by Kiralan et al. (2017) concluded that hexanal and (E)-2-heptenal were the major VOC formed during the oxidation of BSO at 60 °C.